arxiv:1804.10622v1 [astro-ph.ga] 27 apr 2018 · we report alma long-baseline observations of orion...
TRANSCRIPT
Draft version May 1, 2018Typeset using LATEX twocolumn style in AASTeX61
A KEPLERIAN DISK AROUND ORION SrcI, A ∼ 15 M� YSO
Adam Ginsburg,1 John Bally,2 Ciriaco Goddi,3, 4 Richard Plambeck,5 and Melvyn Wright5
1Jansky fellow of the National Radio Astronomy Observatory, 1003 Lopezville Rd, Socorro, NM 87801 USA2CASA, University of Colorado, 389-UCB, Boulder, CO 803093ALLEGRO/Leiden Observatory, Leiden University, PO Box 9513, 2300 RA Leiden, the Netherlands4Department of Astrophysics/IMAPP, Radboud University Nijmegen, PO Box 9010, 6500 GL Nijmegen, the Netherlands5Radio Astronomy Laboratory, University of California, Berkeley, CA 94720
ABSTRACT
We report ALMA long-baseline observations of Orion Source I (SrcI) with resolution 0.03-0.06′′ (12-24 AU) at 1.3
and 3.2 mm. We detect both continuum and spectral line emission from SrcI’s disk. We also detect a central weakly
resolved source that we interpret as a hot spot in the inner disk, which may indicate the presence of a binary system.
The high angular resolution and sensitivity of these observations allow us to measure the outer envelope of the rotation
curve of the H2O 55,0 − 64,3 line, which gives a mass MI ≈ 15 ± 2 M�. We detected several other lines that more
closely trace the disk, but were unable to identify their parent species. Using centroid-of-channel methods on these
other lines, we infer a similar mass. These measurements solidify SrcI as a genuine high-mass protostar system and
support the theory that SrcI and the Becklin Neugebauer Object were ejected from the dynamical decay of a multiple
star system ∼500 years ago, an event that also launched the explosive molecular outflow in Orion.
Corresponding author: Adam Ginsburg
[email protected]; [email protected]
arX
iv:1
804.
1062
2v1
[as
tro-
ph.G
A]
27
Apr
201
8
2
1. INTRODUCTION
Orion Source I (SrcI) is the closest candidate form-
ing high-mass (M > 8 M�) star, and as such is the
most important protostar for testing basic theories of
how massive stars form. However, despite its relative
proximity at a mere ≈ 415 pc from the sun (Menten
et al. 2007; Kim et al. 2008), the mass of SrcI has been
the subject of prolonged debate, with several estimates
putting its mass below the classic 8 M� threshold for a
single star to go supernova (Heger et al. 2003).
Several attempts have been made to measure the mass
of Orion SrcI using the rotation curve of various molec-
ular lines:
• Kim et al. (2008) used 3D VLBI measurements of
SiO masers to infer a source mass M = 8M�.
• Matthews et al. (2010) used 3D VLBI measure-
ments of SiO masers to infer a mass M ≈ 8 − 10
M�.
• Hirota et al. (2014) observed H2O emission from
the v2 = 0, 102,9 − 93,6 and v2 = 1, 52,3 − 61,6lines. They made velocity centroid maps of the
position of peak intensity as a function of veloc-
ity to measure the rotationally supported mass in
SrcI. They obtained a mass estimate of 5 − 7 M�using a model of a simple uniform ring orbiting at
Keplerian velocity.
• Plambeck & Wright (2016) measured both the
continuum SED and the rotation curve of gas
around SrcI. They used a centroiding and model-
ing approach similar to Hirota et al. (2014) to mea-
sure the rotation curve of SiO and vibrationally
excited CO and infer the source mass M ∼ 5 − 7
M�.
• Hirota et al. (2017) used Si18O J=12-11 to infer a
mass M = 8.7 using a similar approach to Hirota
et al. (2014) and Plambeck & Wright (2016).
SrcI’s mass is an important parameter in models of
the origin of the Orion Outflow. Several authors argue
that SrcI, BN, and SrcN (or, alternatively, SrcX; Luh-
man et al. 2017) were part of a single non-hierarchical
multiple system that underwent dynamical decay, and
this decay somehow triggered the outflow (Bally & Zin-
necker 2005; Rodrıguez et al. 2005; Goddi et al. 2011;
Moeckel & Goddi 2012; Bally et al. 2011, 2015, 2017;
Rodrıguez et al. 2017). However, others have noted that
the lower masses inferred for SrcI above are incompati-
ble with this scenario (Chatterjee & Tan 2012; Plambeck
& Wright 2016; Farias & Tan 2017), which requires a
mass MI & 15 M�. An alternative scenario is described
in which BN was ejected from the Trapezium and had
a close encounter with SrcI that triggered the outflow
(Tan 2008a,b; Chatterjee & Tan 2012). A third alter-
native, that the outflow is driven by many independent
sources (Beuther & Nissen 2008), is disfavored by the
overall symmetry of the outflow (Bally et al. 2017).
We present new measurements of SrcI’s mass, finding
it has MI ∼ 15 M�. In Section 2, we present details of
the observations. In Section 3, we discuss measurements
of the continuum and spectral lines. Section 4 discusses
these results and some of their simple physical impli-
cations. We conclude in Section 5. Several appendices
present additional figures and detailed method discus-
sion.
2. OBSERVATIONS
Observations were taken with two configurations in
each of Band 3, 6, and 7 at ALMA as part of project
2016.1.00165.S. The epochs and broad details about the
configuration are given in Table 1. The multiconfigura-
tion data were combined for all images considered here.
The flux and phase calibrators are listed in Table 1.
Band 7 data are not discussed in this work because the
data for the long-baseline observations were not deliv-
ered by the time of submission; we record the observa-
tional details here for completeness since they are part
of the same project in the ALMA archive.
Continuum images were produced with several weight-
ing parameters to emphasize different scales, though
most of the discussion here will be limited to the robust
-2 weighted images with the highest resolution. The
calibrated data delivered from the ALMA QA2 process
were imaged directly, since we found that self-calibration
did not improve the image; we suspect the unmodeled,
resolved-out emission prevents us from obtaining good
calibration solutions.
To emphasize the disk scales and eliminate ripple arti-
facts produced by poorly-sampled large-scale structure
in the map, we used data only from baselines > 150 m
(115 kλ, angular scales < 1.8′′) in the robust -2 images
used for disk fitting and modeling.
The continuum images have dynamic range in the
vicinity of SrcI of about 200-400. These values are re-
ported in Table 2 and are measured by taking the ratio
of the peak intensity in SrcI to the standard deviation
within a neighboring, apparently signal-free region (an
r = 0.7′′ circle 1.4′′ to the northwest of SrcI).
Spectral line image cubes were produced covering the
complete data set to identify lines associated with the
disk. Cubes were produced centered on SrcI with ro-
bust 0.5 and -2 weightings. These cubes were only
3
Table 1. Observation Summary
Date Band Array Observation Duration Baseline Length Range # of antennae FluxCal PhaseCal
seconds meters
08-Oct-2016 6 12m 2332 17-3144 43 J0522-3627 J0541-0541
31-Oct-2016 7 12m 2671 19-1124 42 J0522-3627 J0532-0307
19-Sep-2017 6 12m 5556 41-12147 42 J0522-3627 J0541-0541
24-Sep-2017 3 12m 5146 21-12147 41 J0423-0120 J0541-0541
25-Sep-2017 3 12m 5180 41-14854 42 J0423-0120 J0541-0541
Table 2. Continuum Image Parameters
Band Robust Beam Major Beam Minor Beam PA TB/Sν RMS Source I Sν,max Dynamic Range
′′ ′′ ◦ 103 K Jy−1 mJy beam−1 mJy beam−1
B6 -2 0.037 0.022 67.0 30.3 0.087 19.638 220
B3 -2 0.065 0.041 50.9 53.2 0.038 14.179 370
cleaned in the 0.5′′× 0.5′′ region immediately surround-
ing SrcI, therefore lines with significant emission from
the surrounding medium may be significantly affected
by sidelobes. Continuum-subtracted cubes were pro-
duced by subtracting the median across the 1.8 GHz
bandwidth in each spectral window. All cube anal-
ysis was performed using spectral-cube (https://
spectral-cube.readthedocs.io/en/latest/).
Relevant parameters of the cubes are described in Ta-
ble 3. For the noise estimate, we use the median ab-
solute deviation (MAD) to estimate the standard devi-
ation over the full continuum-subtracted cutout cube,
which effectively ignores the few channels that have sig-
nificant line emission (the directly-measured standard
deviation and MAD-estimated standard deviation differ
by < 5%).
Cutouts of the data used for the analysis in this work
along with the software and scripts used for the anal-
ysis are presented at https://zenodo.org/record/
1213350.
3. RESULTS
3.1. Continuum
We detect the disk in the continuum at 3.2 mm, 1.3
mm, and 0.8 mm (Figure 1 shows the 3.2 and 1.3 mm
images1). At 1.3 mm, where we have enough resolution
to clearly distinguish the line-emitting region from the
disk midplane, we detect spectral lines only from the
surfaces above and below the continuum disk (Figure
1At the time of submission, the long-baseline 0.8 mm data prod-ucts had not been delivered, so they are excluded from the anal-ysis presented here.
2). The nondetection of lines in the disk midplane is a
strong indication that the continuum is optically thick,
as has previously been noted (e.g., Plambeck & Wright
2016).
We fit the highest-resolution 1.3 mm and 3.2 mm con-
tinuum image with a simple model to determine the ba-
sic observational structure. The optimization was per-
formed using a Levenberg-Marquardt fitter (Newville
et al. 2014). We used a linear model (i.e., an infinitely
thin perfectly edge-on disk) for the disk, with endpoints
and amplitude as free parameters. This simple model
left significant residuals, so we added a two-dimensional
Gaussian smoothing kernel as another three free param-
eters to obtain a substantially better fit. The models
and their residuals are shown in Appendix A.
We determined that the disk is resolved in both di-
rections, with a vertical FWHM height of about 20 AU
and a length of about 100 AU (Table 4). These mea-
surements are close to those published by Plambeck &
Wright (2016), though their data only marginally re-
solved the source at wavelengths 1.3 mm and shorter.
This simple model leaves a significant residual com-
pact source near the center of the disk, which we mea-
sured by adding a smeared point source to the model (see
Appendix A). We have allowed the source to be smeared
only in the direction of the disk’s elongation, requiring
only two additional free parameters. This source is dis-
cussed further in Section 4.6.
Table 4 lists the fitted parameters. It includes mea-
surements of the total integrated intensity recovered in
the model and the ratio of the compact central source to
the total. We also display fits to the Reid et al. (2007)
4
Table 3. Line Cube Parameters
Band SPW Freq. Range Robust Beam Major Beam Minor Beam PA RMS RMS Channel Width
GHz ′′ ′′ ◦ mJy beam−1 K km s−1
B3 0 85.463-87.337 -2 0.066 0.043 45.0 2.5 144.8 3.4
B3 1 87.358-89.232 -2 0.064 0.050 41.5 2.5 121.1 3.3
B3 2 97.462-99.336 -2 0.061 0.039 45.8 2.0 108.4 3.0
B3 3 99.358-101.232 -2 0.060 0.038 45.3 2.3 122.2 2.9
B6 0 229.168-231.042 -2 0.026 0.022 64.0 2.8 115.9 1.3
B6 1 231.835-233.709 -2 0.026 0.021 61.6 3.1 125.0 1.3
B6 2 214.277-216.151 -2 0.027 0.023 62.8 3.2 131.5 1.4
B6 3 216.976-218.850 -2 0.030 0.023 55.2 3.2 120.5 1.3
B3 0 85.463-87.337 0.5 0.101 0.072 40.6 0.8 17.3 3.4
B3 1 87.358-89.232 0.5 0.098 0.080 40.3 0.7 15.0 3.3
B3 2 97.462-99.336 0.5 0.091 0.060 43.6 0.7 16.8 3.0
B3 3 99.358-101.232 0.5 0.081 0.058 39.6 0.7 18.2 2.9
B6 0 229.168-231.042 0.5 0.043 0.035 -88.1 0.9 14.6 1.3
B6 1 231.835-233.709 0.5 0.043 0.034 -87.4 1.0 15.7 1.3
B6 2 214.277-216.151 0.5 0.046 0.037 -88.7 1.2 18.2 1.4
B6 3 216.976-218.850 0.5 0.049 0.039 72.7 1.0 14.1 1.3
7 mm continuum data with the same model; these fits
do not contain any absolute astrometry information.
The disk position angle points to within 2 degrees of
the Becklin-Neugebauer object (Orion BN); the PA of
the vector from SrcI to Source BN is -37.6 degrees, while
the measured disk position angle is -36 to -37 degrees.
This coincidence was noted by Bally et al. (2011) and
Goddi et al. (2011).
The disk has a peak brightness temperature at 1.3 mm
of ∼ 600 K at the position of the compact source and
∼ 400−500 K at other positions, with a gradual decline
from the center to the exterior. These measurements
agree with the continuum model of Plambeck & Wright
(2016), who inferred the presence of an optically thick
T = 500 K surface from the SED. The 3.2 mm contin-
uum has a higher peak brightness temperature at the
position of the central compact source, but otherwise is
consistent with the 1.3 mm brightness (see Figure 1).
3.2. SiO Lines
We detect several SiO lines, including the 1 mm 28SiO
v=0 and v=1 J=5-4 lines, the the 3 mm isotopologue
lines 29SiO v=0 and v=1 J=2-1, and the 3 mm 28SiO
v=0 and v=1 J=2-1 lines. Several of these transitions
are known and well-studied masers. Some images of
these data are shown in Appendix B, but because the
emphasis of this work is not on the outflow, we do not
discuss the SiO further here.
3.3. Water Line
The next brightest line after the masing SiO lines
is the H2O 55,0 − 64,3 line at 232.68670 GHz, with
EU = 3461.9 K. Hirota et al. (2012) detected this line in
2′′ resolution ALMA Science Verification data, but be-
lieved it to be masing. We report here that, because it
is similar in morphology and excitation level to the 336
GHz vibrationally excited water line reported in Hirota
et al. (2014), and it has a peak brightness temperature
∼ 1500± 100 K (in the robust -2 maps; see Figure 2), it
is most likely a thermal line.The water line traces an X-shaped feature above and
below the disk, resembling the overall distribution of
SiO masers. The water is not directly aligned with the
continuum disk (Figures 2 and 19), but it does exhibit
emission parallel to the disk at small (< 20 AU) sep-
aration. The morphology of this line confirms that it
traces both the disk and the inner rotating outflow dis-
cussed by Hirota et al. (2017) (see also Kim et al. 2008;
Matthews et al. 2010).
Because the water emission is thermal, it exhibits less
extreme brightness fluctuations across the image than
the SiO masers, allowing us to fit an upper-envelope
velocity curve in Section 3.5. The kinematics of the
water line away from the disk are discussed further in
Appendix C.
3.4. Other lines
5
Table 4. Continuum Fit Parameters
Frequency Disk FWHM Disk Radius Disk PA Pt RA Pt Dec Pt Amp Pt Width Pt Flux Total Flux Pt %
GHz AU AU ◦ s ′′ mJy AU mJy mJy
43.165 25 ± 0.44 41 ± 1.1 -36 - - 1.00 ± 0.01 16 ± 0.69 3 10 29%
93.3 17 ± 0.09 37 ± 0.71 -38 0.518 -0.0405 2.00 ± 0.03 9.4 ± 0.38 5.7 57 10%
224.0 21 51 ± 0.78 -37 0.518 -0.0409 3.20 23 15 280 5.5%
The pointlike source position is given as RA seconds and Dec arcseconds offset from ICRS 5h35m14s -5d22m30s. The error onthis position is 0.003s (RA) and 0.0003′′ (Dec). For the 7 mm data, the position is left blank because we do not have
astrometric information for those data (they were self-calibrated on a bright maser whose position was not well-constrained).The disk FWHM is the vertical full-width half-maximum of the fitted Gaussian profile. No formal parameter errors were
measured for several of the 224.0 GHz fitted quantities because of a linear algebra failure in the fitter; the errors are likelysimilar to the 93.3 GHz fit errors. The Pt Flux and Total Flux columns report the integrals of the best-fit models.
0.2 0.1 0.0 0.1 0.2RA offset (")
0.2
0.1
0.0
0.1
0.2
Dec
offs
et ("
)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
S 3m
m [m
Jy b
eam
1 ]
5
0
5
10
15
20
25
30
35
40
T B [K
]
(a)
0.2 0.1 0.0 0.1 0.2RA offset (")
0.2
0.1
0.0
0.1
0.2
Dec
offs
et ("
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
S 1.3
mm
[mJy
bea
m1 ]
5
0
5
10
15
20
25
30
35
40
T B [K
]
(b)
Figure 1. The robust -2 continuum image of Orion SrcI at 3.2 mm (left) and 1.3 mm (right). The beam is shown in thebottom left, with size 0.065′′ × 0.041′′ at PA= 50.9◦ (3.2 mm, left) and 0.037′′ × 0.022′′ at PA= 67◦ (1.3 mm, right). Contoursare overlaid at TB =50, 100, 150, 200, 300, 400, and 500 K. The displayed coordinates are offsets from ICRS 05:35:14.5172-05:22:30.612 (3.2 mm) and ICRS 05:35:14.5173 -05:22:30.6135 (1.3 mm).
Several unidentified lines are observed in emission at
the outer edge of the continuum disk. A table of their
approximate rest frequencies is presented in Appendix
D. They all share a common morphology, though they
vary in strength. The peak signal from these lines ap-
pears around the TB ∼ 150 K contour in the robust -2
weighted 1.3 mm continuum images (Figure 2), and the
lines are particularly strong at the endpoints of the disk.
Little line emission is detected where the continuum is
brightest, TB & 300 K.
The best explanation for these lines is that they trace
the outer surface of a mostly optically thick (in the
continuum) disk. In this scenario, the lines have an
excitation temperature similar to the brightness tem-
perature of the disk, but have optical depths of order
τ ∼ 0.1−1. Directly toward the disk continuum emission
peak, since the line excitation temperature is the same
as the background continuum temperature, Tex = Tbg,
no emission (or absorption) is observed. Just above and
below the disk midplane continuum peak, the dust col-
umn density (and therefore optical depth) drops rapidly,
but the molecular optical depth drops more slowly, so
some emission is still observed (Tex ∼ 200 − 500 K,
but τline ∼ 0.1 − 0.5, resulting in the TB,max ∼ 100
K observed). At the disk endpoints, the column density
of molecular material is higher because we are looking
along the tangent of the disk, so the line optical depth
and therefore brightness are greater.
Since these lines only appear immediately around the
disk, and in particular because they peak just outside of
the dust emission along the disk axis, they are the most
direct tracers of the disk’s kinematics.
3.5. Kinematics: a Keplerian disk
6
0.2 0.1 0.0 0.1 0.2RA offset (")
0.2
0.1
0.0
0.1
0.2
Dec
offs
et ("
)
20
40
60
80
100
120
140
160
T B [K
](a)
0.2 0.1 0.0 0.1 0.2RA offset (")
0.2
0.1
0.0
0.1
0.2
Dec
offs
et ("
)
200
400
600
800
1000
1200
T B [K
]
(b)
0.2 0.1 0.0 0.1 0.2RA offset (")
0.2
0.1
0.0
0.1
0.2
Dec
offs
et ("
)
200
400
600
800
1000
1200
1400
T B [K
]
(c)
Figure 2. Peak intensity map of an unknown line (U230.322; left) and the H2O line (middle, right) with continuum overlaidin red contours at levels of 50, 300, and 500 K. White contours are shown at 50, 100, and 150 K (left), and 500, 750, 1000, and1250 K (middle, right). The H2O and unknown line clearly trace different physical structures, as they exhibit no coincidentemission peaks. The water line does not exhibit emission directly along the disk midplane. The left two figures are robust 0.5weighted images, while the right is a robust -2 weighted image with higher resolution and poorer sensitivity. The U230.322 lineis not detected in the robust -2 cubes. The positions shown are offsets from coordinate ICRS 05:35:14.5184 -05:22:30.6194.
The disk appears to exhibit a Keplerian rotation
curve, which allows us to use the velocity profile to mea-
sure the central mass. Following Seifried et al. (2016),
we measure the outer edge of the detected emission in
a position-velocity (PV) diagram to define the rotation
curve surrounding SrcI. The emission along each line-of-
sight in the PV diagram is followed from its peak down
to some emission threshold. We use a threshold of 5-σ,
as was done in Seifried et al. (2016), but we also assess
the importance of this threshold in Section 3.5.2. Ad-
ditionally, to facilitate direct comparison with previous
works, we use the centroid-of-velocity-channel approach
in Appendix E (although Seifried et al. (2016) warn it
may underestimate the central mass) and obtain similar
results, albeit from different spectral lines.
Of the detected lines, the H2O line spans the widest
range of velocities as a function of radius. As shown inFigure 3, there is H2O emission spanning at least radii
r ≈ 10 to 100 AU. Many other molecules, most of which
we have not been able to identify, span a range of radii
30-80 AU, while SiS spans 30 AU to an unconstrained
outer radius. The SiO lines span varying radii, but they
are predominantly detected far from the disk midplane
(however, see Appendix B, in which SiO also exhibits
Keplerian velocity curves).
We find that a 15 M� edge-on Keplerian rotation
curve fits2 the outer edge of the H2O line well (Figure
3). The 15 M� profile is also an acceptable match to the
outer profiles of the unidentified lines described in Sec-
2We do not report best-fit parameters and statistical errors herebecause the errors on the outer envelope are poorly character-ized and likely dominated by systematic errors such as channeldiscretization.
tion 3.4 and shown in supplemental figures in Appendix
F.
A lower-mass central source is consistent with the H2O
data only if we allow for substantial line broadening
driven by turbulence. We estimate an upper limit on
the turbulent line broadening of FWHM . 4 km s−1
based on the narrowest features observed in the H2O and
other unknown lines, which results in a one-sided broad-
ening of HWHM < 2 km s−1; if such line-broadening
is present, the mass may be lower by ∼ 20− 30%. How-
ever, since Flaherty et al. (2017) observe stringent upper
limits on turbulence in lower-mass disks, we expect tur-
bulent broadening to be relatively small, and possibly
negligible. The upper-limit line broadening we observe
is both consistent with line blending from unresolved
kinematics within the beam and is close to the intrinsic
velocity resolution of our data.
A substantially smaller mass, such as the 5-10 M�suggested previously (Plambeck & Wright 2016; Hirota
et al. 2014), is inconsistent with the data: for models
with such masses, emission is clearly detected outside of
the predicted Keplerian curve (see, e.g., the green curve
in Figure 3).
3.5.1. Examination of alternative velocity profile models
To support the argument that the velocity profiles are
Keplerian, as opposed to some other power-law profile
as has been found for the outer envelopes of several low-
mass YSOs (Lee et al. 2017; Aso et al. 2017; Ohashi
et al. 2014; Lindberg et al. 2014; Murillo et al. 2013), we
show power-law fits to the outer envelope velocity pro-
file of the H2O line in Figure 4. This figure convincingly
demonstrates that a power law α = 1 (e.g., as observed
in the outer parts of low-mass YSO disks; Aso et al.
7
-0.2 -0.1 0.0 0.1 0.2
40
20
0
-20
-40
Offset ["]
V LSR
[km
/s]
50 au
0.000
0.005
0.010
0.015
0.020
0.025
S [J
y be
am1 ]
Figure 3. Position-velocity diagram of H2O 55,0−64,3. The colorbars show average intensity along the extracted region in unitsof mJy beam−1. The blue line with dots is the outer envelope of the velocity curve determined using the method of Seifriedet al. (2016). The red solid and green dotted curves show the Keplerian velocity profile surrounding a 15 and 10 M� centralsource, respectively. White dashed lines indicate the adopted source central position 05h35m14.5172s -05d22m30.618s (ICRS)and central velocity (5.5 km s−1). The purple dashed lines show the full orbital path for radii of 10 and 100 AU and indicatethe approximate limits of the disk. This PV diagram is extracted from the midplane of the robust 0.5 image, but the emissiondisplayed is beam-smeared from just above and below the continuum disk.
8
2017) is inconsistent with the data. While the best-fit
profile of α ≈ 0.4 is slightly shallower than the Keplerian
α = 0.5 curve, both shallow profiles are consistent with
the data. The limited spectral resolution of our data
is evident in this plot, where there is little separation
in velocity from ∼ 30 − 50 AU; higher spectral resolu-
tion observations or more sophisticated modeling may
be able to provide a tighter constraint on the power-law
slope.
0 10 20 30 40 50 60 70 80 90Offset from center (AU)
12
14
16
18
20
22
24
26
28
30
Offs
et fr
om c
entro
id v
eloc
ity (k
m s
1 )
= 0.42= 1= 0.5
Figure 4. Radial profile of the outer-envelope velocity pro-file extracted from Figure 3b. The red and blue points rep-resent the redshifted and blueshifted components of the ve-locity profile, respectively. The velocities are shown relativeto the best-fit centroid velocity for the H2O line, vLSR = 5.2km s−1. The curves show the best-fit power-law (black solidline) and the best-fit curves with fixed powerlaw indices of0.5 (magenta dotted line) and 1 (green dashed line). Themagenta filled curve shows a powerlaw index α = 0.5, i.e., aKeplerian rotation curve, for the range 13M� < M < 17M�.
The mass for the α = 0.5 curve is M = 15 M�. We do
not determine masses for the other models since they are
not consistent with a pointlike gravitational potential.
We show the curves for a central 13-17 M� source in
filled magenta: since there are many points above the
curve, a more massive central source is plausible, while
a less massive source is unlikely.
3.5.2. An estimate of the error on the mass measurement
We assess the uncertainty introduced by the threshold
level adopted in the velocity envelope profile measure-
ment. Figure 5 shows the effects of increasing or de-
creasing the threshold, which is to decrease or increase
the measured mass, respectively. These figures suggest
that our measurement uncertainty with the PV envelope
fitting technique is approximately 2 M�.
4. DISCUSSION
4.1. The mass of SrcI
We measure a mass for the object at the center of
the disk of MI = 15 M�, which is higher than most
measurements previously reported. Our mass measure-
ment is higher than previous works in part because our
spatial resolution is high enough to allow a direct fit of
the rotation curve to the outer envelope of an emission
line in position-velocity space. Additionally, though, the
greater sensitivity of these observations allowed us to
detect the outer envelope of the H2O position-velocity
diagram and detect - and resolve - several unknown lines
that directly trace the disk. The inconsistency between
these new estimates and the lower masses previously de-
rived from SiO measurements hints that, in this system,
SiO chemically selects a kinemetically distinct region
from the disk.
Even if SrcI consists of an equal-mass binary, this mass
measurement confirms that the Orion Molecular Cloud
is presently a region with ongoing high-mass star forma-
tion.
4.2. The luminosity of SrcI
Since we observe an optically thick surface, we can in-
fer the luminosity required to keep such a surface at the
observed TB,1.3mm ≈500 K assuming it is heated only
by radiation. Taking the disk radius to be 50 AU, the
required central source luminosity is 6500 L�. This es-
timate should be taken as a lower limit, since the inner
disk is likely to be optically thick and capable of shield-
ing the outer disk, thereby keeping the observed τ = 1
surface at 1.3 mm cooler than would be produced by
radiative equilibrium with the central star.
4.3. Properties of the disk
Our observations yield disk properties nearly identi-
cal to those in Plambeck & Wright (2016), so we do not
revisit their disk mass or density estimates. We note,
however, that these new observations have sufficient an-
gular resolution to distinguish the molecular lines that
trace the outflow from those that directly trace the disk.
4.4. The dynamical decay scenario
Several authors (Gomez et al. 2008; Goddi et al. 2011;
Bally et al. 2011) suggested that the high proper mo-
tion of SrcI, BN, and SrcN, combined with the observed
H2 outflow, implied the outflow and the runaway stars
were produced in the same single event ∼ 500 years
ago. That event was the dynamical decay of a non-
hierarchical multiple system, i.e., it was the interaction
9
0 20 40 60 80Offset from center (AU)
12
14
16
18
20
22
24
26
28
30Of
fset
from
cen
troid
vel
ocity
(km
s1 )
0 20 40 60 80Offset from center (AU)
12
14
16
18
20
22
24
26
28
30
Offs
et fr
om c
entro
id v
eloc
ity (k
m s
1 )
0 20 40 60 80Offset from center (AU)
12
14
16
18
20
22
24
26
28
30
Offs
et fr
om c
entro
id v
eloc
ity (k
m s
1 )
Figure 5. Demonstration of the effect of a changing threshold in the Seifried method. The panels show a lower 3-σ threshold(left), the adopted 5-σ threshold (center), and a higher 7-σ threshold (right). The magenta highlighted region is the same 13-17M� Keplerian curve shown in Figure 4.
of multiple stars at the center of a small cluster. More
recent observations by Luhman et al. (2017) have shown
that SrcN is unlikely to have participated in this inter-
action, but instead that SrcX, another star within the
same field, has high proper motion that points back to
the interaction center (Bally et al, in prep).
Farias & Tan (2017) report that, while any dynami-
cal decay scenario involving Sources I, X, and BN that
can reproduce the observed proper motions are unlikely,
those with a higher mass for SrcI (MI > 14 M�) arethe only ones capable of producing the observed proper
motions3. Our observed higher mass for SrcI, MI & 15
M�, therefore implies that the dynamical interaction
scenario remains viable.
4.5. Is the disk consistent with the dynamical ejection
model?
Plambeck & Wright (2016) argue that both the mass
of SrcI and the presence of the disk rule out the dy-
namical ejection model of Bally et al. (2011). We have
shown that the star is significantly more massive, but
what about the disk?
3Their results are similar to those obtained in Goddi et al. (2011)and Moeckel & Goddi (2012), but now with SrcX instead of SrcNas the third member of the interaction.
Following Bally et al. (2011), we note that the disk
truncates at R < 50 AU. At this radius, the orbital
timescale is ∼ 70 years, so gas at the disk’s outer radius
would have had five to ten dynamical times to relax into
a circular disk configuration after the explosive event.
The alignment of SrcI’s disk with the I-BN vector is
consistent with a dynamical interaction between these
sources. If the ejection resulted in SrcI and BN being
launched in nearly opposite directions from their center
of mass (which must have been moving in the rest frame
of the Orion nebula; Bally et al. in prep), any material
around SrcI that remained bound would be dragged in
the direction of SrcI, and would therefore have a result-
ing angular momentum vector orthogonal to the direc-
tion of motion. Any material with velocity relative to
SrcI
v < vesc = 23km s−1(MI/15M�)1/2(r/50 AU)−1/2
would remain bound. Assuming SrcI’s present-day
proper motion of 11.5 km s−1 reflects its velocity at
the time of ejection, less than half of the original disk
mass would have been lost, while the rest would remain
bound (material moving in the direction directly oppo-
site SrcI’s ejection direction would have net velocity rel-
ative to SrcI high enough to escape; the greatest mass
loss would occur if the disk was already in the direction
of SrcI’s eventual launch). Material outside R & 200
10
AU (where vI = 11.5 km s−1 > vesc) would likely all
have become unbound, while other material would be
retained in a disk parallel to Source I’s proper motion.
4.6. The compact source in the disk
We have detected a compact (but marginally resolved)
source near the center of the disk at both 3.2 mm and
1.3 mm. The source has a spectral energy distribution
that is shallow from 3.2 mm to 1.3 mm (α ≈ 1.1). Since
the source is nearly coincident with an edge-on disk that
we show is optically thick at 1.3 mm, it is likely that the
source is thermal but is significantly attenuated by the
disk at 1.3 mm and seen with less attenuation at 3.2
mm.
Reid et al. (2007) used comparable-resolution 7 mm
VLA data to infer the presence of a 2.2 mJy source at
the center of the SrcI disk. The compact-source-to-disk
flux ratio at 7 mm was ∼ 30%, substantially higher than
we observe at 1.3 mm and somewhat higher than at 3.2
mm (Table 4). The spectral index of this compact source
from 7 to 3.2 mm is α = 1.6, approaching that of an
optically thick blackbody.
The central source has a surface temperature T ≥1250 K, the brightness temperature of a 2.2 mJy source
within a 41 × 28 milliarcsecond beam at 7 mm. If the
source is a 5000 K spherical blackbody (e.g., Testi et al.
2010), it must have a radius R = 7.5 AU. Such a gigantic
star is implausible, as it would produce a luminosity of
1.5×106 L�, several orders of magnitude higher than
the total luminosity in the region. We therefore argue
that this emission source is not a star.
What is the emission mechanism from this central
source? It could simply be hot, optically thick dust
that is partly obscured by the cooler disk at higher fre-
quencies. The extension of this ‘source’ along the disk
direction (Appendix A, Figure 6) suggests that we are
seeing the hot inner disk. As pointed out by Plambeck &
Wright (2016), it is quite unlikely to be classical free-free
emission from protons and electrons, since there are no
detected recombination lines. However, it is still plau-
sible that the emission is produced by brehmsstrahlung
emission from HI and H2 (Reid et al. 2007; Baez-Rubio
et al. 2018).
The source is slightly offset from the center of the disk
by 5.8 ± 1.5 AU in projection4. This offset, combined
with the source’s extent, implies that it is not a single
central source, but instead is a hot region of the inner
4We measure the errors on the source position by fitting a 2DGaussian model to an image with the disk model subtracted.While this approach yields a useful statistical error, it does notaccount for the systematic error introduced by fitting the diskmodel.
disk. Such an asymmetry in the disk could be driven
either by instability in the disk or, if the central star is
a binary, by the proximity of the more luminous com-
panion.
If this source is an inner edge of the disk, it may imply
the presence of a binary that has cleared the area within
r < 6−10 AU. Since a tight binary is one of the expected
outcomes of the dynamical interaction scenario (Goddi
et al. 2011), this detection of the inner region in dust
emission provides additional circumstantial evidence for
that scenario.
If SrcI’s central source is a binary, and the measured
offset of ∼ 5 AU between the disk midpoint and the
central emission source is real, we can guess that the
binary’s orbit is . 5 AU. For such an orbital radius, the
orbital timescale is only ∼ 3 years. It will therefore be
productive to re-observe SrcI over the next several years
to see if the hot spot moves on such a timescale.
5. CONCLUSIONS
We report observations that resolve SrcI’s disk in both
continuum and line emission. We measure the mass of
SrcI by fitting the rotation curve with a Keplerian disk
model, finding the following:
1. The central source has mass M = 15 ± 2 M�,
where the the error bar represents the range of
consistent models rather than a typical 1 − σ sta-
tistical uncertainty.
2. The H2O 55,0 − 64,3 line is not masing and kine-
matically traces both the upper envelope of the
disk and the lower portion of the outflow.
3. We observe several lines that trace the disk kine-
matics directly, though the molecules producing
these lines remain unidentified. These lines are
visible only toward the outskirts of the disk and
are morphologically distinct from both the H2O
and SiO lines that follow the outflow.
4. A compact source in the approximate center of
the disk is resolved at 1.3 mm, and it is slightly
off-center. It therefore is most likely a hot region
of the inner disk. It may be produced by time-
varying illumination from an unequal mass binary.
The mass we have measured is higher than in several
recent publications because both the resolution and sen-
sivity of our observations were greater. These new data
allowed us to identify and measure the spectral features
that directly trace the disk kinematics, while previous
data convolved the disk and outflow kinematics. This
11
higher measured mass implies that the dynamical decay
scenario for the SrcI- BN - SrcX system is viable.
We thank the anonymous referee for a thorough and
helpful review. This paper makes use of the follow-
ing ALMA data: ADS/JAO.ALMA#2016.1.00165.S
ALMA is a partnership of ESO (representing its mem-
ber states), NSF (USA) and NINS (Japan), together
with NRC (Canada), MOST and ASIAA (Taiwan), and
KASI (Republic of Korea), in cooperation with the
Republic of Chile. The Joint ALMA Observatory is op-
erated by ESO, AUI/NRAO and NAOJ. The National
Radio Astronomy Observatory is a facility of the Na-
tional Science Foundation operated under cooperative
agreement by Associated Universities, Inc.
Software: The software used to make this ver-
sion of the paper is available from github at https://
github.com/keflavich/Orion_ALMA_2016.1.00165.S
(https://doi.org/10.5281/zenodo.1181877) with
hash 68a49ad(2018-04-26). The tools used include
spectral-cube (https://doi.org/10.5281/zenodo.
591639 and https://github.com/radio-astro-tools/
spectral-cube) and radio-beam (https://github.
com/radio-astro-tools/radio-beam, https://doi.
org/10.5281/zenodo.1181879) from the radio-astro-tools
package ( radio-astro-tools.github.io), astropy
(Astropy Collaboration et al. 2013), astroquery
(astroquery.readthedocs.io, https://doi.org/10.
5281/zenodo.591669 ) and CASA (McMullin et al.
2007).
REFERENCES
Aso, Y., Ohashi, N., Aikawa, Y., et al. 2017, ApJ, 849, 56
Astropy Collaboration, Robitaille, T. P., Tollerud, E. J.,
et al. 2013, A&A, 558, A33
Baez-Rubio, A., Jimenez-Serra, I., Martın-Pintado, J.,
Zhang, Q., & Curiel, S. 2018, ApJ, 853, 4
Bally, J., Cunningham, N. J., Moeckel, N., et al. 2011, ApJ,
727, 113
Bally, J., Ginsburg, A., Arce, H., et al. 2017, ApJ, 837, 60
Bally, J., Ginsburg, A., Silvia, D., & Youngblood, A. 2015,
A&A, 579, A130
Bally, J., & Zinnecker, H. 2005, AJ, 129, 2281
Beuther, H., & Nissen, H. D. 2008, ApJL, 679, L121
Chatterjee, S., & Tan, J. C. 2012, ApJ, 754, 152
Farias, J. P., & Tan, J. C. 2017, ArXiv e-prints,
arXiv:1712.08485
Flaherty, K. M., Hughes, A. M., Rose, S. C., et al. 2017,
ApJ, 843, 150
Goddi, C., Humphreys, E. M. L., Greenhill, L. J.,
Chandler, C. J., & Matthews, L. D. 2011, ApJ, 728, 15
Gomez, L., Rodrıguez, L. F., Loinard, L., et al. 2008, ApJ,
685, 333
Heger, A., Fryer, C. L., Woosley, S. E., Langer, N., &
Hartmann, D. H. 2003, ApJ, 591, 288
Hirota, T., Kim, M. K., & Honma, M. 2012, ApJL, 757, L1
Hirota, T., Kim, M. K., Kurono, Y., & Honma, M. 2014,
ApJL, 782, L28
Hirota, T., Machida, M. N., Matsushita, Y., et al. 2017,
ArXiv e-prints, arXiv:1712.04606
Kim, M. K., Hirota, T., Honma, M., et al. 2008, PASJ, 60,
991
Lee, C.-F., Li, Z.-Y., Ho, P. T. P., et al. 2017, ApJ, 843, 27
Lindberg, J. E., Jørgensen, J. K., Brinch, C., et al. 2014,
A&A, 566, A74
Luhman, K. L., Robberto, M., Tan, J. C., et al. 2017,
ApJL, 838, L3
Matthews, L. D., Greenhill, L. J., Goddi, C., et al. 2010,
ApJ, 708, 80
McMullin, J. P., Waters, B., Schiebel, D., Young, W., &
Golap, K. 2007, in Astronomical Society of the Pacific
Conference Series, Vol. 376, Astronomical Data Analysis
Software and Systems XVI, ed. R. A. Shaw, F. Hill, &
D. J. Bell, 127
Menten, K. M., Reid, M. J., Forbrich, J., & Brunthaler, A.
2007, A&A, 474, 515
Moeckel, N., & Goddi, C. 2012, MNRAS, 419, 1390
Murillo, N. M., Lai, S.-P., Bruderer, S., Harsono, D., & van
Dishoeck, E. F. 2013, A&A, 560, A103
Newville, M., Stensitzki, T., Allen, D. B., & Ingargiola, A.
2014, LMFIT: Non-Linear Least-Square Minimization
and Curve-Fitting for Python¶, , ,
doi:10.5281/zenodo.11813.
https://doi.org/10.5281/zenodo.11813
Ohashi, N., Saigo, K., Aso, Y., et al. 2014, ApJ, 796, 131
Plambeck, R. L., & Wright, M. C. H. 2016, ApJ, 833, 219
Reid, M. J., Menten, K. M., Greenhill, L. J., & Chandler,
C. J. 2007, ApJ, 664, 950
Rodrıguez, L. F., Dzib, S. A., Loinard, L., et al. 2017, ApJ,
834, 140
Rodrıguez, L. F., Poveda, A., Lizano, S., & Allen, C. 2005,
ApJL, 627, L65
Seifried, D., Sanchez-Monge, A., Walch, S., & Banerjee, R.
2016, MNRAS, 459, 1892
12
Tan, J. C. 2008a, in Astronomical Society of the Pacific
Conference Series, Vol. 387, Massive Star Formation:
Observations Confront Theory, ed. H. Beuther, H. Linz,
& T. Henning, 346
Tan, J. C. 2008b, ArXiv e-prints, arXiv:0807.3771
Testi, L., Tan, J. C., & Palla, F. 2010, A&A, 522, A44
13
APPENDIX
A. CONTINUUM MODELING FIGURES
In this appendix, we show figures illustrating the continuum modeling process. Figures 6 and 7 show the model and
residuals for the band 6 and band 3 data, highlighting the significantly improved fit as more model parameters are
added.
In Figure 6, the apparent point sources at the edge of the disk in the central column are artificial features introduced
by the model; since the model is forced to be smooth, the best-fit model is one that is more centrally peaked, which
results in an under-prediction of the disk brightness toward the edges.
Also in this figure, there appears to be a faint ‘halo’ of emission at the ∼ 30 K level around the modeled region. The
halo is asymmetric, with a greater extent toward the southwest. If this feature is not an artifact of the data reduction,
which we cannot rule out, it is likely to be from optically thin dust above and below the disk, since it is not detected
in the 3.2 mm data.
B. THE SIO OUTFLOW IN PV SPACE
To illustrate the change in velocity structure with height from the disk, we show position-velocity diagrams of SiO
v=0 J=5-4 in Figure 8 and 29SiO J=5-4 in Figure 9. These images are extracted from equal distances above and below
the disk midplane. At greater distances from the midplane, the high-velocity, low-separation features fade out, while
more low-velocity material becomes visible at larger separations. These structures are similar to what was shown in
the 484 GHz Si18O J=12-11 line in Figure 2 of Hirota et al. (2017). Our data are consistent with their interpretation
that the SiO isotopologues trace a rotating, expanding outflow. In the innermost slice, which shows the SiO emission
that just skirts the edges of the disk, the velocity curve is consistent with the 15 M� Keplerian curve overlaid.
C. A DEEPER EXAMINATION OF THE WATER LINE: EVIDENCE THAT IT TRACES THE DISK
KINEMATICS
The H2O-derived mass presented in Section 3.5 relies on the H2O line tracing the disk kinematics. Since the H2O
clearly also traces the outflow, showing the same X-shaped morphology as the SiO, it does not trace just the disk.
Nonetheless, the midplane position-velocity slice of the H2O line does appear to genuinely trace disk kinematics.
Qualitatively, the PV diagram appears exactly as expected for a disk with an inner and out radial cutoff.
To assess possible contamination from the outflow, we compare position velocity slices at different vertical displace-
ments from the disk center in Figure 10. The left panel shows the kinematic signature we attribute to the disk, which
closely resembles that predicted for a pure Keplerian rotation curve. In contrast, the middle panel is likely dominated
by outflow emission, since it shows material 0.05-0.1′′ (20-40 AU) above the disk, i.e., just outside the 1-σ height of the
continuum disk. While the outflow continues to show some motion similar to that of the disk, it lacks the characteristic
convex shape of a Keplerian orbit at higher velocities and separation. Finally, the rightmost panel shows that the
water emission nearly disappears at heights h > 0.1′′ = 40 AU while the maximum velocities observed get smaller
(dv < 10 km s−1), suggesting that rotation slows in the outflow.
Figure 10 also characterizes some of the ‘forbidden’ velocity components, i.e., those seen in quadrants 2 and 4. These
components get stronger at higher vertical positions on the disk, implying that they come from the outflow, not the
disk. The “ring” shape observed in the high-latitude figures indicates the outflow is expanding (see, e.g., the model in
supplementary figure 1 of Hirota et al. 2017). The velocity asymmetry, which shows an excess toward the red side of
the disk and outflow, is also present in SiO. We do not have a straightforward explanation for this asymmetry except
to assert that it implies an asymmetry in the direction of mass ejection in the outflow. These velocity components are
unlikely to be produced by infall motions, since they are observed perpendicular to the disk along the direction of the
outflow.
D. STACKED SPECTRA
We reported the detection of several unidentified lines. To measure their frequencies precisely, we performed a
stacking analysis in which we adopt the velocity field of the U232.511 line, shift all spectra across the disk to the
same velocity frame, and average them. We stacked the robust 0.5 cubes, as the surface brightness sensitivity of the
robust -2 cubes was too poor to justify stacking. We then fit the lines with Gaussians to determine their centroid
frequency. We searched within a narrow range of velocities (vLSR = 3 − 8 km s−1) for known lines in the Splatalogue
14
0100200300400500
T B [K
]
0100200300400500
T B [K
]
40302010
01020304050
T B [K
]
Figure 6. A series of plots showing the band 6 continuum models used and their residuals. The top row shows the models,starting from a simple 1D linear model convolved with the beam (left), continuing with a disk smoothed with a broader beamto account for scale height (middle), and finally a version of the middle model with a smeared point source added (right). Thefit parameters are given in Table 4. The second and third row show the residuals (data - model) for each of the models in thetop row; the bottom row uses a narrow linear scale to emphasize the lower-amplitude residuals, while the top two use an arcsinhstretch to display the full dynamic range.
collection of line catalogs using astroquery. While many of the lines have plausible carriers within 1-2 km s−1, such
as highly-excited CH3OCHO or variants of SO2, there is no consistent pattern to the detected lines and no individual
species can explain more than a few of the observed lines. These disk-averaged spectra are shown in Figure 12 with
the lines labeled.
We list the line frequencies (which we use as line names), fitted Gaussian widths, and fitted amplitudes from the
stacked spectra in Table 5.
E. DISK PARAMETER DETERMINATION METHOD COMPARISON
To compare fairly with Hirota et al. (2014) and Plambeck & Wright (2016), we used the centroid-velocity method
to measure the central source mass. In this approach, we fit two-dimensional Gaussian profiles to each ‘blob’ in each
velocity channel in the PPV cubes of spectral lines. Unlike previous works, we have had to fit multiple Gaussians in
several channels, since we resolve the structure and see ‘blobs’ both above and below the disk. Figures 13, 14, and 15
show the results of this analysis.
15
0100200300400500
T B [K
]
0100200300400500
T B [K
]
40302010
01020304050
T B [K
]
Figure 7. A series of plots showing the band 3 continuum models used and their residuals. See the caption of Figure 6 fordetails.
-0.2 0.0 0.2
40
30
20
10
0
-10
-20
-30
Offset ["]
V LSR
[km
/s]
50 au0.000
0.005
0.010
0.015
0.020
S [J
y be
am1 ]
(a)
-0.2 0.0 0.2
40
30
20
10
0
-10
-20
-30
Offset ["]
V LSR
[km
/s]
50 au0.000
0.005
0.010
0.015
0.020
S [J
y be
am1 ]
(b)
-0.2 0.0 0.2
40
30
20
10
0
-10
-20
-30
Offset ["]
V LSR
[km
/s]
50 au0.0025
0.0000
0.0025
0.0050
0.0075
0.0100
0.0125
0.0150
S [J
y be
am1 ]
(c)
-0.2 0.0 0.2
40
30
20
10
0
-10
-20
-30
Offset ["]
V LSR
[km
/s]
50 au 0.002
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
S [J
y be
am1 ]
(d)
Figure 8. Position-velocity slices of SiO v=0 J=5-4 along the disk direction at four heights: (a) |h| < 0.05′′, (b) 0.05′′ < |h| <0.1′′, (c) 0.1′′ < |h| < 0.15′′, (d) 0.15′′ < |h| < 0.2′′. Contours are overlaid at 5 and 10 σ. These images are produced fromthe robust -2 weighted cubes. The missing emission around v = 0 km s−1 is likely caused by image filtering effects; at thesevelocities, there is extended, smooth SiO emission from the surrounding cloud.
16
-0.2 0.0 0.2
40
30
20
10
0
-10
-20
-30
Offset ["]
V LSR
[km
/s]
50 au0.000
0.005
0.010
0.015
0.020
S [J
y be
am1 ]
(a)
-0.2 0.0 0.2
40
30
20
10
0
-10
-20
-30
Offset ["]
V LSR
[km
/s]
50 au0.000
0.005
0.010
0.015
0.020
S [J
y be
am1 ]
(b)
-0.2 0.0 0.2
40
30
20
10
0
-10
-20
-30
Offset ["]
V LSR
[km
/s]
50 au0.000
0.005
0.010
0.015
0.020
S [J
y be
am1 ]
(c)
-0.2 0.0 0.2
40
30
20
10
0
-10
-20
-30
Offset ["]
V LSR
[km
/s]
50 au0.000
0.005
0.010
0.015
0.020
S [J
y be
am1 ]
(d)
Figure 9. Position-velocity slices of 29SiO v=0 J=5-4 along the disk direction at four heights: (a) |h| < 0.05′′, (b) 0.05′′ < |h| <0.1′′, (c) 0.1′′ < |h| < 0.15′′, (d) 0.15′′ < |h| < 0.2′′. Contours are overlaid at 5, 10, 15, 20, and 25 σ. These images are producedfrom the robust -2 weighted cubes. While similar to the 28SiO shown in Figure 8, there is a remarkable position-velocity ringat high elevations that is coincident with many of the SiO and H2O masers.
-0.2 0.0 0.2
40
30
20
10
0
-10
-20
-30
Offset ["]
V LSR
[km
/s]
50 au0.0025
0.0000
0.0025
0.0050
0.0075
0.0100
0.0125
0.0150
0.0175
S [J
y be
am1 ]
(a)
-0.2 0.0 0.2
40
30
20
10
0
-10
-20
-30
Offset ["]
V LSR
[km
/s]
50 au0.000
0.005
0.010
0.015
0.020
S [J
y be
am1 ]
(b)
-0.2 0.0 0.2
40
30
20
10
0
-10
-20
-30
Offset ["]
V LSR
[km
/s]
50 au 0.002
0.000
0.002
0.004
0.006
0.008
0.010
0.012
S [J
y be
am1 ]
(c)
Figure 10. Position-velocity slices of the H2O 55,0 − 64,3 line at different heights above the disk plane from the robust -2 datacube. The left panel shows the inner 0.1′′ (i.e., near the midplane, |h| < 0.05′′, |h| < 20 AU), the middle shows the range0.05′′ < |h| < 0.1′′, and the right shows the range 0.1′′ < |h| < 0.15′′. All three panels show averages over the specified range,with the colorbar showing intensity in mJy beam−1. Contours are overlaid at 5, 10, 15, and 20 σ. The red solid lines showthe Keplerian profile for a 15 M� central source, and the purple dashed lines show the orbital track for a particle at 30 and 80AU for such a source; these are included primarily to guide the eye. The middle and right panel are dominated by the outflow,while the left panel is dominated by the Keplerian orbital profile.
We have modeled the velocity profile assuming an edge-on, uniform, optically thin disk with a sharp central hole and
outer truncation. Position-velocity curves derived with this approach are shown in the above figures. Figure 16 shows
the curves for a range of masses and radii. This model approach is the same used by Plambeck & Wright (2016). We
fit this model to the centroid data points. The fits were performed on the average positional offset at each velocity,
since for many velocities there were two or more Gaussian components fitted in the image. The mass, inner and outerradius, and centroid velocity were left as free parameters. The fit results are shown in the legend of Figures 13 and
14; in Figure 15, we show only a fiducial model because the best-fit model did not describe the data well.
The positions of the fitted components are significantly different for each species, which helps illustrate why previous
estimates of SrcI’s mass were low. Fits to both the SiO line in Figure 14 and water in Figure 15 have lower masses
than the fit to the U232.511 line in Figure 13 that more closely traces the disk.
In the edge-on disk models, different inner-radius cutoffs have an effect on the inner velocity profile slope similar to
changing the central mass, so it is likely that the disk parameters, rather than the central source mass, dominate our
uncertainties in this approach. Figure 16 demonstrates this effect: the inner slope of a 5 M�, 20 AU < r < 50 AU disk
is indistinguishable from a 20 M�, 30 AU < r < 80 AU disk, though the latter extends to higher velocity and radius.
E.1. A demonstration of issues with the centroid-of-velocity method
Figure 17 shows an example of how the centroid-of-velocity approach produces lower mass fits for some lines,
particularly H2O and SiO. The figure shows both the optically thin edge-on disk model and the midplane-extracted
position-velocity diagram with overlaid centroid fits. The centroid fits notably do not extend nearly as far as emission
is visible. This discrepancy results from the midplane emission being much fainter than some of the off-plane emission.
17
-0.2 -0.1 0.0 0.1 0.2
40
20
0
-20
-40
Offset ["]
V LSR
[km
/s]
50 au
0.000
0.005
0.010
0.015
0.020
0.025
0.030
S [J
y be
am1 ]
(a)
-0.2 -0.1 0.0 0.1 0.2
40
20
0
-20
-40
Offset ["]
V LSR
[km
/s]
50 au
0.000
0.005
0.010
0.015
0.020
0.025
0.030
S [J
y be
am1 ]
(b)
Figure 11. Duplicate of Figure 3 for robust -2 data with different vertical extents included in the position-velocity slice.The left figure shows the average PV slice over the vertical range range h = ±0.05′′, and the right right shows the same withh = ±0.1′′.
By contrast, a similar side-by-side comparison of the edge-on optically thin model with the U232.511 line reveals
a better match. In Figure 18, the overall structure of the observed position-velocity diagram is well-matched to the
model.
F. ADDITIONAL FIGURES SHOWING THE DISK
We include several additional figures showing the disk moment 0 maps and position-velocity profiles for some other
unknown lines. These figures show that the lines displayed in the main text are not unique.
Figure 19 shows moment 0 maps, which provide a slightly different view from the peak intensity maps shown in
Figure 2. Figures 20 and 21 show peak intensity, moment 0, and position-velocity maps of the U232.511 and U217.980
lines.
18
229.5 230.0 230.5 231.0Frequency (GHz)
0.010
0.005
0.000
0.005
0.010
Jy
bea
m
U2
29
.24
7
Si34
S 1
3-1
2
U2
29
.55
0
U2
29
.68
2
U2
29
.81
9
U2
30
.32
2
12C
O 2
-1
U2
30
.72
6
U2
30
.78
0
U2
30
.96
6
(a)
232.0 232.5 233.0 233.5Frequency (GHz)
0.010
0.005
0.000
0.005
0.010
Jy
bea
m
H3
0α
U2
32
.16
3
U2
32
.51
1
U2
32
.63
4
H2O
v2=
1 5
5,0−
64,
3
U2
33
.17
1
U2
33
.60
8
(b)
214.5 215.0 215.5 216.0Frequency (GHz)
0.010
0.005
0.000
0.005
0.010
Jy
bea
m
29SiO
v=
0 J=
5-4
U2
14
.54
9
U2
14
.63
7
U2
14
.74
2
U2
14
.94
0
U2
15
.00
9
SiO
v=
1 J=
5-4
13C
H3O
H 4
2,2−
31,
2
(c)
217.0 217.5 218.0 218.5Frequency (GHz)
0.010
0.005
0.000
0.005
0.010
Jy
bea
m
SiO
v=
0 J=
5-4
U2
17
.22
9
U2
17
.54
7
U2
17
.66
6
SiS
12
-11
U2
17
.78
0
HC
3N
24
-23
U2
18
.58
4
(d)
Figure 12. Plots of the stacked spectra from spectral windows 0, 1, 2, and 3 (see Table 3) with detected lines labeled. Thespectra are shown with the same y-axis limits; bright SiO and H2O emission is cut off. In spectral window 2, the region around215.5-215.6 GHz, near the SiO v=1 J=5-4 maser line (which is the brightest line we detect) is affected by imaging artifacts fromthe cleaning process.
19
Table 5. Unknown Line Frequencies
Line Name Frequency Fitted Width Fitted Amplitude
GHz km s−1 mJy
U214.549 214.549 4.1 0.7
U214.637 214.637 3.2 0.3
U214.742 214.742 - -
U214.940 214.940 4.3 4.6
U215.009 215.009 4.6 2.7
U217.229 217.229 2.6 1.3
U217.547 217.547 6.8 1.2
U217.666 217.666 7.2 1.1
U217.980 217.980 5.2 5.6
U218.584 218.584 4.5 1.6
U229.247 229.247 4.8 5.9
U229.550 229.550 8.0 1.1
U229.682 229.682 15.5 3.0
U229.819 229.819 4.7 1.8
U230.322 230.322 4.7 3.8
U230.726 230.726 5.8 1.5
U230.780 230.780 6.7 5.1
U230.966 230.966 10.4 1.1
U232.163 232.163 4.2 1.5
U232.511 232.511 6.7 6.2
U232.634 232.634 7.8 0.8
U233.171 233.171 4.2 2.3
U233.608 233.608 6.8 1.5
The frequencies listed have a systematic uncertainty of about 2 km s−1 (1.5 MHz) because they are referenced to the U232.511line, which has an unknown rest frequency. The rest frequency used for the U232.511 line was selected to maximize thesymmetry of the emission around 5 km s−1. Some lines were detected in only part of the disk and therefore had bad or
malformed profiles in the stacked spectrum; these have fits marked with -’s.
20
0.0
8
0.0
4
0.0
0
0.0
4
0.0
8
Offset RA (arcsec)
0.10
0.05
0.00
0.05
0.10
Off
set
Dec
(arc
sec)
-0.2 -0.1 0.0 0.1 0.2
-10
0
10
20
30
Offset Position (arcsec)
VLSR [km
s−
1]
M= 12. 6Rin = 37Rout = 57
(a)
0.0
8
0.0
4
0.0
0
0.0
4
0.0
8
Offset RA (arcsec)
0.10
0.05
0.00
0.05
0.10
Off
set
Dec
(arc
sec)
-0.2 -0.1 0.0 0.1 0.2
-10
0
10
20
30
Offset Position (arcsec)
VLSR [km
s−
1]
M= 15. 4Rin = 38Rout = 54
(b)
Figure 13. Results of the centroid-velocity analysis for the U232.511 line (left) and the U230.322 line (right). The left panelshows the locations of fitted centroids in the position-position plane relative to the midpoint of the disk. The position of thecentral compact source is marked with a grey circle at the center. The grey line indicates the disk midplane as determinedfrom the continuum modeling. The circles are colored by their velocity as indicated in the right panel. The right panel showsa position-velocity diagram of these same centroids. The dotted external curves show Keplerian velocity profiles for a 15 M�(red solid) central source; this curve does not represent what should be observed in a centroid-of-velocity plot. The black curveshows the predicted centroid velocity profile of an optically-thin edge-on disk with parameters displayed in the figure.
21
0.2
0
0.1
6
0.1
2
0.0
8
0.0
4
0.0
0
0.0
4
0.0
8
0.1
2
0.1
6
0.2
0
Offset RA (arcsec)
0.20
0.15
0.10
0.05
0.00
0.05
0.10
0.15
0.20O
ffse
t D
ec
(arc
sec)
-0.2 -0.1 0.0 0.1 0.2
-20
0
20
40
Offset Position (arcsec)
VLSR [km
s−
1]
M= 10. 0Rin = 13Rout = 63
Figure 14. Results of the centroid-velocity analysis for the SiO v=1 J=5-4 line. See Figure 13 for details. Note that thedata are inconsistent with the disk model; the SiO emission fitted here traces the bottom of the outflow and possibly somecomponents of the disk upper atmosphere. The average positions at each velocity are closer to a reasonable fit.
22
0.2
0
0.1
6
0.1
2
0.0
8
0.0
4
0.0
0
0.0
4
0.0
8
0.1
2
0.1
6
0.2
0
Offset RA (arcsec)
0.20
0.15
0.10
0.05
0.00
0.05
0.10
0.15
0.20O
ffse
t D
ec
(arc
sec)
-0.2 -0.1 0.0 0.1 0.2
-20
0
20
40
Offset Position (arcsec)
VLSR [km
s−
1]
Figure 15. Results of the centroid-velocity analysis for the H2O line. See Figure 13 for details. The model disk fit to thesedata was a very poor fit, so we have instead overlaid a model that is not a fit to the data with 15.5 M�, rinner = 17 AU,and router = 66 AU in the right panel. Note that the centroid positions do not extend as far from the source center as theunknown lines; this effect is a symptom of the blending of lines of sight in the centroid-based approach, since the H2O line’sfaintest emission can be seen extending to at least as great a distance from the central source as the unknown lines in theposition-velocity diagrams.
60 40 20 0 20 40 60Offset (AU)
30
20
10
0
10
20
30
Vobs [
km s−
1]
M=20, 20 < r < 50M=20, 30 < r < 50M=20, 30 < r < 80M=5, 20 < r < 50M=5, 30 < r < 50M=5, 30 < r < 80
Figure 16. Plots of the predicted centroid rotation curves for different masses and inner and outer radial cutoffs. As discussedin Appendix E, this figure illustrates the ambiguity between the radial extent of the disk and the mass of the central source,since an M = 5 M� central source with a 20 < R < 50 AU disk has the same slope in the inner part as a M = 20 M� sourcewith a 30 < R < 80 AU disk.
23
0.1 0.0 0.1 0.2Offset Position (arcsec)
30
20
10
0
10
20
30
40O
ffse
t D
ec
(arc
sec)
-0.2 -0.1 0.0 0.1 0.2
-20
0
20
40
Offset Position (arcsec)
VLSR [km
s−
1]
Figure 17. The edge-on disk model used to fit the centroid-of-velocity curves with M = 15M�, rin = 25 AU, and rout = 65AU (left) and the centroid-of-velocity measurements overlaid on the midplane position-velocity diagram of the H2O line (right).
24
0.1 0.0 0.1 0.2Offset Position (arcsec)
10
0
10
20
30O
ffse
t D
ec
(arc
sec)
-0.2 -0.1 0.0 0.1 0.2
-10
0
10
20
30
Offset Position (arcsec)
VLSR [km
s−
1]
Figure 18. The edge-on disk model used to fit the centroid-of-velocity curves with M = 15M�, rin = 25 AU, and rout = 65AU (left) and the centroid-of-velocity measurements overlaid on the midplane position-velocity diagram of the U232.511 line(right).
0.2 0.1 0.0 0.1 0.2RA offset (")
0.2
0.1
0.0
0.1
0.2
Dec
offs
et ("
)
500
250
0
250
500
750
1000
1250
T Bdv
[K k
m s
1 ]
(a)
0.2 0.1 0.0 0.1 0.2RA offset (")
0.2
0.1
0.0
0.1
0.2
Dec
offs
et ("
)
0
2500
5000
7500
10000
12500
15000
17500
20000
T Bdv
[K k
m s
1 ]
(b)
0.2 0.1 0.0 0.1 0.2RA offset (")
0.2
0.1
0.0
0.1
0.2
Dec
offs
et ("
)
0
5000
10000
15000T B
dv [K
km
s1 ]
(c)
Figure 19. Moment-0 (integrated intensity) map of the U230.322 (left) and H2O (middle, right) line with continuum overlaidin contours. Continuum contours from the robust -2 map are shown in red at levels of 50, 300, and 500 K. Contours of the linedata are shown at 5, 10, and 15 σ. The left two figures are robust 0.5 weighted images, while the right is a robust -2 weightedimage with higher resolution and poorer sensitivity. The U230.322 line is not detected in the robust -2 cubes.
25
0.2 0.1 0.0 0.1 0.2RA offset (")
0.2
0.1
0.0
0.1
0.2
Dec
offs
et ("
)
500
0
500
1000
1500
2000
T Bdv
[K k
m s
1 ](a)
0.2 0.1 0.0 0.1 0.2RA offset (")
0.2
0.1
0.0
0.1
0.2
Dec
offs
et ("
)
25
50
75
100
125
150
175
200
T B [K
]
(b)
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
40
30
20
10
0
-10
-20
-30
Offset ["]
V LSR
[km
/s]
50 au 0.002
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
S [J
y be
am1 ]
(c)
Figure 20. Moment 0 and peak intensity map of U232.511, similar to Figures 2 and 19. The rightmost panel shows a position-velocity diagram extracted from the disk midplane. The overlaid curve shows the Keplerian velocity profiles for a 15 M� centralsource in red. The dashed magenta lines show the velocity curves at r=30 and 70 AU. In the left and right panels, contours ofthe line data are shown at 5 and 10 σ, and in the center panel, they are at 50, 100, 150, and 200 K.
0.2 0.1 0.0 0.1 0.2RA offset (")
0.2
0.1
0.0
0.1
0.2
Dec
offs
et ("
)
500
0
500
1000
1500
T Bdv
[K k
m s
1 ]
(a)
0.2 0.1 0.0 0.1 0.2RA offset (")
0.2
0.1
0.0
0.1
0.2
Dec
offs
et ("
)
20
40
60
80
100
120
140
160
180
T B [K
]
(b)
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
40
30
20
10
0
-10
-20
-30
Offset ["]
V LSR
[km
/s]
50 au 0.002
0.000
0.002
0.004
0.006
0.008
0.010
0.012
S [J
y be
am1 ]
(c)
Figure 21. Moment 0 and peak intensity map of U217.980, similar to Figures 2 and 19. The rightmost panel shows a position-velocity diagram extracted from the disk midplane. The overlaid curve shows the Keplerian velocity profiles for a 15 M� centralsource in red. The dashed magenta lines show the velocity curves at r=30 and 70 AU. In the left and right panels, contours ofthe line data are shown at 5 and 10 σ, and in the center panel, they are at 50, 100, and 150 K.